[Solved] CS 6810/7810: Wavelets and Wavelet Algorithms Assignment 6 2D Ordered HWT

Learning Objectives1. 2D Ordered HWT2. Applying 2D Ordered HWT to ImagesIntroductionIn this assignment, you will implement the 2D Ordered HWT and apply it to images. For the sake of simplicity, we willassume that input images are square and their length and width are integral powers of 2.

Problem 1Implement a class TwoDHWT.java with the following static method that applies the 2D Ordered HWT to the 2D signal sig for the number of iterations given in the second argument.public static void ordFwdDWTForNumIters(double[][] sig, int num_iters) {}The class WaveletAlgos_S17_HW06.java has several 2D signals de ned. You can use these signals for debugging. Weanalyzed some of these signals in Lecture 8. There is a method testOrd2DHWT(double[][] data, int num_iters) inWaveletAlgos_S17_HW06 to test dierent numbers of iterations. Note that the method TwoDHWT.ordFwdDWTForNumItershas a third boolean argument set to false. You do not have to have this argument. In my implementation, this is adebugging ag. When I want intermediate states displayed, I set this argument to true.public class WaveletAlgos_S17_HW06 {static double[][] signal_2x2_1 = {{9, 7},{5, 3}};static double[][] signal_4x4_1 = {{9,7,6,2},{5,3,4,4},{8,2,4,0},{6,0,2,2}};static double[][] signal_8x8_2 = {{8, 5, 4, 8, 6, 8, 10, 8},{8, 10, 10, 4, 10, 4, 8, 2},{6, 10, 2, 4, 2, 6, 1, 6},{0, 8, 0, 10, 10, 6, 6, 10},{8, 8, 8, 0, 4, 8, 6, 2},{10, 6, 2, 2, 6, 6, 6, 8},{2, 4, 10, 10, 10, 4, 6, 10},{10, 6, 10, 6, 6, 4, 4, 4}};1public static void testOrd2DHWT(double[][] data, int num_iters) {final int dim = data.length;TwoDHWT.ordFwdDWTForNumIters(data, num_iters, false);System.out.println(Result Matrix);Utils.display2DArray(data, dim, dim);System.out.println();}}Let us run a few tests. In the rst test, we appy the 2D HWT to signal_2x2_1.public static void main(String[] args) {testOrd2DHWT(signal_2x2_1, 1);}Here is the output.Result Matrix6.0 1.02.0 0.0In the second test, we apply the HWT to signal_4x4_1 for one iteration.public static void main(String[] args) {testOrd2DHWT(signal_4x4_1, 1);}Here is the output.Result Matrix6.0 4.0 1.0 1.04.0 2.0 3.0 1.02.0 0.0 0.0 1.01.0 0.0 0.0 1.0We now run 2 iterations of 2D HWT on the same signal.public static void main(String[] args) {testOrd2DHWT(signal_4x4_1, 2);}Here is the output.Result Matrix4.0 1.0 1.0 1.01.0 0.0 3.0 1.02.0 0.0 0.0 1.01.0 0.0 0.0 1.0In the third test, let us apply the 2D HWT to the 88 signal analyzed in lecture 8 de ned as signal_8x8_2.public static void main(String[] args) {testOrd2DHWT(signal_8x8_2, 1);}Here is the output after one scale.Result Matrix7.75 6.5 7.0 7.0 0.25 0.5 1.0 2.06.0 4.0 6.0 5.75 -3.0 -3.0 0.0 -2.258.0 3.0 6.0 5.5 1.0 2.0 -1.0 0.55.5 9.0 6.0 6.0 0.5 1.0 2.0 -1.0-1.25 -0.5 0.0 2.0 1.25 -2.5 -2.0 -1.02.0 -1.0 -2.0 -2.25 1.0 2.0 -2.0 -0.250.0 1.0 0.0 -1.5 -1.0 2.0 -1.0 1.5-2.5 1.0 1.0 2.0 -1.5 -1.0 1.0 -1.02We now apply the 2D Ordered HWT to the signal for two iterations.public static void main(String[] args) {testOrd2DHWT(signal_8x8_2, 2);}Here is the output.Result Matrix6.0625 6.4375 0.8125 0.0625 0.25 0.5 1.0 2.06.375 5.875 0.375 0.125 -3.0 -3.0 0.0 -2.251.0625 0.5625 -0.1875 -0.0625 1.0 2.0 -1.0 0.5-0.875 -0.125 2.125 0.125 0.5 1.0 2.0 -1.0-1.25 -0.5 0.0 2.0 1.25 -2.5 -2.0 -1.02.0 -1.0 -2.0 -2.25 1.0 2.0 -2.0 -0.250.0 1.0 0.0 -1.5 -1.0 2.0 -1.0 1.5-2.5 1.0 1.0 2.0 -1.5 -1.0 1.0 -1.0Finally, we apply the 2D Ordered HWT to the signal for three iterations.public static void main(String[] args) {testOrd2DHWT(signal_8x8_2, 3);}Here is the output.Result Matrix6.1875 0.03125 0.8125 0.0625 0.25 0.5 1.0 2.00.0625 -0.21875 0.375 0.125 -3.0 -3.0 0.0 -2.251.0625 0.5625 -0.1875 -0.0625 1.0 2.0 -1.0 0.5-0.875 -0.125 2.125 0.125 0.5 1.0 2.0 -1.0-1.25 -0.5 0.0 2.0 1.25 -2.5 -2.0 -1.02.0 -1.0 -2.0 -2.25 1.0 2.0 -2.0 -0.250.0 1.0 0.0 -1.5 -1.0 2.0 -1.0 1.5-2.5 1.0 1.0 2.0 -1.5 -1.0 1.0 -1.0Problem 2Implement the method in WaveletAlgos_S17_HW06.java:pubic static void createScaledImageOf2DHWT(input_img_path, output_img_path, num_iters)This method applies the speci ed number of iterations of the 2D Ordered HWT to the image in input_img_path andsaves the result in the image in output_img_path. Consider the image in gure 1.Figure 1: Ornament 1Then the following main() generates the images displayed in gures 2, 3, 4, and 5.public static main(String[] args) {createScaledImageOf2DHWT(ornament_01.jpg, ornament_01_1_scale.jpg, 1);createScaledImageOf2DHWT(ornament_01.jpg, ornament_01_2_scales.jpg, 2);createScaledImageOf2DHWT(ornament_01.jpg, ornament_01_3_scales.jpg, 3);createScaledImageOf2DHWT(ornament_01.jpg, ornament_01_4_scales.jpg, 4);}3Figure 2: Ornament 1: 1 scale of 2D HWT saved in ornament 01 1 scale.jpgThe top left square matrix in gure 2 is the coarser (or downsampled) representation of the original signal after 1 iteration.The top right square matrix in each image is the horizontal wavelets. The bottom left matrix is the vertical wavelets.The bottom right matrix is the horizontal wavelets.If 2 iterations/scales are applied to the image in gure 1, the 2nd scale is applied to the top left matrix. This top leftmatrix is replaced by four smaller matrices, as shown in gure 3, where the top left matrix is the coarser image, the topright matrix contains the horizontal wavelets after the 2nd scale, the bottom left matrix contains the vertical waveletsafter the 2nd scale, and the bottom right matrix contains the diagonal wavelets after the 2nd scale.Figure 3: Ornament 1: 2 scales of 2D HWT saved ornament 01 2 scales.jpgThe same recursive pattern continues in gures 4 and 5.Figure 4: Ornament 1: 3 scales of 2D HWT saved in ornament 01 3 scales.jpgFigure 5: Ornament 1: 4 scales of 2D HWT in ornament 01 4 scales.jpgDisplaying the top left images is straightforward in that they are smaller versions of the original image. All we need todo is to convert the doubles into integers. Displaying wavelets is a bit trickier in that some of them will be negative and4no image format, to the best of my knowledge, can handle negative real pixel values. What I did is in the above gures is to scale the wavelets to lie in [0, 255].